Machine Learning Monitor

Problem

Once a Machine Learning model is deployed into the wild its performance can deteriorate when it encounters new data. ML engineers lack a signal that will tell them how models are performing and when they need to be updated.

Approach

The team created a customer-focused dashboard and performed usability tests and heuristic evaluations with Azure Machine Learning Studio Engineers.

Solution

Users can set up model monitoring jobs at deployment time via config files and commands. A detailed dashboard includes features such as: model performance statistics, model comparison, alert configuration, and report generation.

View Poster

Back to Projects

Share